Abstract: Aiming at the problem that the key information in the text is ignored and the classification accuracy is not high, a weighted word2vec CNN and ATT-BiGRU mixed neural network sentiment analysis model is proposed. Since the word vector generated by word2vec cannot highlight the role of text keywords, the term frequency-inverse document frequency (TF-IDF) algorithm is introduced to calculate the vocabulary weight value. Then, the weighted operation word vector is input into the mixed model of CNN and ATT-BiGRU to extract the hidden features. The proposed model extracts text features by Convolutional Neural Network (CNN) and attention-based Bidirectional Gated Recurrent Unit (ATT-BiGRU) to improve text representation. Compared with other algorithms, the results show that the classification accuracy of the proposed model is the highest and the cost is small.